It is shown that discrete-time quantum walks can be used to digitize, i.e., to time discretize fermionic models of continuous-time lattice gauge theory. The resulting discrete-time dynamics is thus not only manifestly unitary, but also ultralocal, i.e., the particle's speed is upper bounded, as in standard relativistic quantum field theories. The lattice chiral symmetry of staggered fermions, which corresponds to a translational invariance, is lost after the requirement of ultralocality of the evolution; this fact is an instance of Meyer's 1996 no-go results stating that no nontrivial scalar quantum cellular automaton can be translationally invariant [D. A. Meyer, J. Stat. Phys. 85, 551 (1996); Phys. Lett. A 223, 337 (1996)]. All results are presented in a single-particle framework and for a (1+1)-dimensional space-time.
Discrete time crystals (DTCs) refer to a novel many-body steady state that spontaneously breaks the discrete time-translational symmetry in a periodically driven quantum system. Here, we study DTCs in a Bose-Einstein condensate bouncing resonantly on an oscillating mirror, using a two-mode model derived from a standard quantum field theory. We investigate the validity of this model and apply it to study the long-time behavior of our system. A wide variety of initial states based on two Wannier modes are considered. We find that in previous studies the investigated phenomena in the evolution time window (
⪅2000 driving periods) are actually ``short-time'' transient behavior though DTC formation signaled by the subharmonic responses is still shown if the interboson interaction is strong enough. After a much longer (about 20 times) evolution time, initial states with no ``long-range'' correlations relax to a steady state, where time-symmetry breaking can be unambiguously defined. Quantum revivals also eventually occur. This long-time behavior can be understood via the many-body Floquet quasieigenenergy spectrum of the two-mode model. A symmetry-breaking edge for DTC formation appears in the spectrum for strong enough interaction, where all quasieigenstates below the edge are symmetry breaking while those above the edge are symmetric. The late-time steady state's time-translational symmetry depends solely on whether the initial energy is above or below the symmetry-breaking edge. A phase diagram showing regions of symmetry-broken and symmetric phases for differing initial energies and interaction strengths is presented. We find that, according to this two-mode model, the discrete time crystal survives for times out to at least
250000 driving periods.
We consider the
O(3) nonlinear sigma model with the $\ensuremath{\theta}$ term and its linear counterpart in 1+1D. The model has discrete time-reflection and space-reflection symmetries at any $\ensuremath{\theta}$, and enjoys the periodicity in $\ensuremath{\theta}\ensuremath{\rightarrow}\ensuremath{\theta}+2\ensuremath{\pi}$. At $\ensuremath{\theta}=0,\ensuremath{\pi}$ it also has a charge-conjugation
C symmetry. Gauging the discrete space-time reflection symmetries is interpreted as putting the theory on the nonorientable
RP2 manifold, after which the $2\ensuremath{\pi}$ periodicity of $\ensuremath{\theta}$ and the
C symmetry at $\ensuremath{\theta}=\ensuremath{\pi}$ are lost. We interpret this observation as a mixed 't Hooft anomaly among charge-conjugation
C, parity
P, and time-reversal
T symmetries when $\ensuremath{\theta}=\ensuremath{\pi}$. Anomaly matching implies that in this case the ground state cannot be trivially gapped, as long as
C,P, and
T are all good symmetries of the theory. We make several consistency checks with various semiclassical regimes, and with the exactly solvable XYZ model. We interpret this anomaly as an anomaly of the corresponding spin-half chains with translational symmetry, parity, and time reversal [but not involving the SO(3)-spin symmetry], requiring that the ground state is never trivially gapped, even if SO(3) spin symmetry is explicitly and completely broken. We also consider generalizations to $\mathbb{C}{P}^{N\ensuremath{-}1}$ models and show that the $C\text{\ensuremath{-}}P\text{\ensuremath{-}}T$ anomaly exists for even
N.